functions to convert empty strings to NA and vice versa

NAMESPACE

@@ -4,6 +4,7 @@ S3method(plot,tidygtfs)
 S3method(print,tidygtfs)
 S3method(summary,tidygtfs)
 export(cluster_stops)
+export(empty_strings_to_na)
 export(filter_feed_by_area)
 export(filter_feed_by_date)
 export(filter_feed_by_stops)

R/io.R

@@ -62,6 +62,9 @@ write_gtfs <- function(gtfs_obj, zipfile, compression_level = 9, as_dir = FALSE)
   # convert sf tables
   gtfs_out = sf_as_tbl(gtfs_obj)
 
+  # convert NA to empty strings
+  gtfs_out <- na_to_empty_strings(gtfs_out)
+
   # data.tables
   gtfs_out <- gtfs_out[names(gtfs_out) != "."]
   gtfs_out <- lapply(gtfs_out, as.data.table)

R/utils.R

@@ -6,3 +6,42 @@ feed_contains <- function(gtfs_obj, table_name) {
   exists(table_name, where = gtfs_obj) ||
     (exists(".", where = gtfs_obj) && exists(table_name, where = gtfs_obj$.))
 }
+
+#' Convert empty strings ("") to NA values in gtfs tables
+#' 
+#' @param gtfs_obj tidygtfs object
+#' 
+#' @return a gtfs_obj where all empty strings in tables have been replaced with NA
+#' 
+#' @export
+empty_strings_to_na = function(gtfs_obj) {
+  tbl_names = names(gtfs_obj)
+  tbl_names <- tbl_names[tbl_names != "."]
+  for(tbl in tbl_names) {
+    if(inherits(gtfs_obj[[tbl]], "data.frame")) {
+      gtfs_obj[[tbl]][gtfs_obj[[tbl]] == ""] <- NA
+    }
+  }
+  gtfs_obj
+}
+
+#' Convert NA values to empty strings ("")
+#'  
+#' @param gtfs_obj tidygtfs object
+na_to_empty_strings = function(gtfs_obj) {
+  lapply(gtfs_obj, function(df) {
+    if(inherits(df, "data.frame")) {
+      df2 = lapply(df, function(.col) {
+        if(is.character(.col)) {
+          .col[is.na(.col)] <- ""
+        }
+        .col
+      })
+      attributes(df2) <- attributes(df)
+      df <- df2
+    } else if(inherits(df, "list")) {
+      df <- na_to_empty_strings(df)
+    }
+    df
+  })
+}

tests/testthat/test-utils.R

@@ -100,3 +100,21 @@ test_that("gtfs_meta", { # empty test
   expect_equal(gtfs_meta, get_gtfs_meta())
 })
 
+test_that("empty_strings_to_na", {
+  gpath = system.file("extdata", "sample-feed-fixed.zip", package = "tidytransit")
+  g1 = read_gtfs(gpath)
+  g_na = empty_strings_to_na(g1)
+  g2 = na_to_empty_strings(g_na)
+
+  for(tbl in names(g1)) {
+    expect_equal(g1[[tbl]], g2[[tbl]])
+  }
+
+  tmppath = tempfile(fileext = ".zip")
+  write_gtfs(g_na, tmppath)
+  g3 = read_gtfs(tmppath)
+
+  for(tbl in names(g1)) {
+    expect_equal(g1[[tbl]], g3[[tbl]])
+  } 
+})